Quickstart: Run a Batch¶

This quickstart drives the simulated reference plant used throughout these docs: Site plant-01, a bulk drug substance plant that runs entirely in software. Its upstream process cell, fermentation-cell, holds the three units you will work with:

fermenter-1 — a 20,000 L stirred-tank bioreactor
seed-tank-1 — a 1,000 L inoculum growth vessel
media-prep-1 — a 5,000 L media batching and sterilisation tank

Downstream, a second cell (recovery-cell) holds a centrifuge and a chromatography skid. There is no physical hardware anywhere in it, so you can drive real recipes against realistic process dynamics from the browser.

You'll need access to a running Cloud-Native DCS instance with this reference plant deployed. If you're standing one up yourself, see Deploy Your Own, which installs the chart and applies exactly this hierarchy. Otherwise your platform team provides the URL and login. Everything below is already deployed. You just drive it.

The install run end to end in one terminal: a kind cluster created from nothing, registry login and pull secret, cert-manager, helm install of the DCS chart from the OCI registry, seven workloads reaching Running, the fermentation hierarchy applied to a Ready site, and the gateway's health endpoint answering healthy on its NodePort — Deploy Your Own, run end to end from an empty cluster to the reference plant. It closes on the gateway answering healthy with five of five unit runtimes up, which is the state this walkthrough starts from.

The five steps below are the rest of that arc. Here they are on the same plant, and every click in the clip is an instruction from this page.

The quickstart run end to end: the fermenter-cip-v1 master recipe walked Approve, Release and Make Effective in the System endpoint, the fermentation cell opened in the HMI, a batch created through + New and started from its row command bar, the row settling on Running with its ISA-88 command set, the fermenter-1 unit dashboard with its control-module cards while the clean-in-place runs, and the completed production record with its phase history in the Data endpoint — Steps 1 through 5, run on the reference plant. The clean-in-place takes about two minutes to run, and the clip marks on screen the stretch of waiting it skips.

This walkthrough touches the Web UI's role-aligned endpoints (/system, /hmi, /data). Each step calls out which endpoint you're in.

1. Find a recipe in the System endpoint¶

Recipes are engineering artifacts. You author and release them in /system, then execute them in /hmi.

Open /system and in the sidebar expand Recipe Library → Master Recipes. The Site dropdown at the top of the recipes page points at the site that contains the reference plant. RBAC confines each login to the sites it has been granted.

What's in the reference plant?

The reference plant lives on Site plant-01 under Enterprise bioworks, with areas upstream and downstream. Pick that site in the Site dropdown if it isn't already selected.

The reference plant ships three master recipes: bds-v2 and bds-v3, two versions of the production fermentation recipe, and fermenter-cip-v1, the clean-in-place turnaround that runs between lots. Every recipe starts in Draft. Promote one to Effective before it can run.

System /system → Master Recipes list with product, version, and target cell for the plant-01 recipes

For the fastest path, pick fermenter-cip-v1. It is a single-unit cycle on the fermenter (fill with rinse water, recirculate, drain), and it runs to completion in about two minutes, where a production fermentation runs for days. Open it and walk it through the approval workflow (Approve → Release → Make Effective) to land it in Effective. Each of the three asks you for a reason before it commits, because each one is an electronic signature: the reason, the signer and a digest of the recipe go into the audit trail together. All three need the dcs-supervisor or dcs-admin role. See Recipes for the full lifecycle, the per-action role table and the CLI/YAML equivalents.

2. Open the process cell in the HMI¶

Switch to /hmi (the operations endpoint). The sidebar carries the site hierarchy, already expanded down to the control modules when only one site is configured. Click the fermentation-cell process cell under plant-01 → upstream, expanding the site node first if it isn't open. The HMI process-cell view shows one card per unit with its current ISA-88 state and, if a batch has allocated the unit, the active unit procedure.

HMI /hmi process cell view: fermenter-1, media-prep-1 and seed-tank-1 as unit cards with their ISA-88 state, and the active unit procedure and batch ID on any unit a batch has allocated

3. Create and start a batch¶

Click Batch in the HMI sidebar to open the Batch Execution view. Click + New, pick the master recipe from step 1, give the batch a unique ID, and submit. The new batch lands in the list in Pending, with Start and Abort as the only commands offered on it.

Click Start on the batch row to kick it off. The batch-operator allocates units and builds the ISA-88 procedural tree (Procedure → Unit Procedure → Operation → Phase), the state goes to Running, and the commands column updates with the ISA-88 commands you can send next (Stop, Hold, Abort, Pause).

HMI Batch Execution: the batch list, with a running batch carrying the Stop, Hold, Abort and Pause commands in its Commands column, and the terminal batches below it carrying none

Prefer the CLI? Authenticate first, and tell the CLI which site holds the reference plant, via the --site flag, the DCS_SITE environment variable, or a config-file default:

dcs login
export DCS_SITE=plant-01   # or pass --site on each command
dcs create batch --recipe fermenter-cip-v1 --batch-id MY-BATCH-001
dcs command Batch MY-BATCH-001 Start

Batch ID and resource name are not the same handle

--batch-id sets the identifier you see in the Batch ID column and on the production record. The Batch resource itself gets a generated name (batch-6jrkm), which dcs create batch prints and dcs get batches carries in its NAME column. dcs command takes either handle.

4. Watch a unit run¶

Back in the HMI, click into fermenter-1. The unit dashboard renders a card for every control module grouped by declared device class (valves, actuators, sensors, controllers, and anything unclassified under other), along with the unit's mode, state, and the active unit procedure when a batch is running on it.

HMI /hmi unit detail: fermenter-1 dashboard with control-module cards grouped by declared device class (valves, actuators, sensors)

Click any card to open its faceplate for its properties, its live tag values, and, for operator-only control moves, its ISA-88 mode switch and manual overrides.

5. Review what happened in the Data endpoint¶

After the batch completes, switch to /data (the production information endpoint). The sidebar views include:

Trends — historical process values for any tag
Audit Trail — every operator action and system event
Batch Records — the consolidated production record for each completed batch (21 CFR Part 11 review sign-off lives here)
Alarm History — past alarm transitions

Data /data → Batch Records: consolidated production records with per-batch state, timing, and review status

Click any completed batch to see its full production record: the frozen recipe snapshot with the parameter values that batch actually ran, and a phase-by-phase execution history giving each phase's procedural path, the unit it ran on, its start and end times, and how it terminated.

That's the whole loop: engineer → operate → review. The rest of the site explains how the pieces work and how to use them for your own process.

Libraries — the control modules, phases, and recipes that ship with the reference plant
Recipes — the master recipe lifecycle (Draft → Approved → Released → Effective)
Batch Execution — ISA-88 state machine and command semantics in depth
Architecture — how the Kubernetes operators and unit runtime fit together
Deploy Your Own — run Cloud-Native DCS on your own cluster